Machine vision and analytics suite boosts safety standards and prevents drowning
One third of all drowning cases occur in pools while lifeguards are on duty. When properly trained and completely attentive, they provide effective protection. However, this is rarely the situation. “When considering large coverage areas, environmental conditions such as sun glare and poor sight angles, distractions like mobile phones and contact with visitors and employees, it’s clear that continuous, effective lifeguarding is nearly impossible,” says Omer Bar-Ilan, coordinator of the EU-funded LYNXIGHT project. The window of opportunity to prevent critical injury during drowning is the first 60 seconds. After that the swimmer loses consciousness and neurological damage begins.
Advanced computer vision and deep learning technology to assist lifeguards
Current systems detect drowning victims only when they lie motionless at the bottom of the pool for a minimum duration, usually about 10 seconds. This solution detects early stages of drowning while the swimmer is still conscious and on the water’s surface. Existing systems only detect drowning. The system’s core is a suite of real-time video analysis algorithms developed by LYNXIGHT. These algorithms detect and classify much more advanced and complex cases, such as an unsupervised child alone in a pool and people playing aggressively. They alert the lifeguard before a potential drowning event further develops. By tracking lifeguards around the pool, the solution monitors activity and measures their routine and performance to ensure they remain alert. Unlike any other system on the market, the innovation tracks all swimmers continuously, simultaneously and anonymously. It collects information about pool usage and swimmer behaviour patterns. This assists management in understanding what goes on in their pool facilities while making operations safer and more efficient. The system can be deployed in any size swimming pool without needing to modify infrastructure. This simple engineering concept offers one or two overhead cameras that can be standard closed-circuit television cameras already installed as part of the pool’s security system. These cameras are networked to a central processing unit. When the system detects and identifies swimmers in distress, it wirelessly alerts the lifeguard or other personnel via a smartwatch. “Being autonomous and self-contained in terms of hardware and installation requirements, the system is easy to set up and maintain, and therefore practical and affordable for any pool operator,” explains Bar-Ilan.
Competitive advantage
The LYNXIGHT team evaluated dozens of existing commercial solutions to identify weaknesses and why they are unsatisfactory in addressing the problem. “This customer validation process has also helped shape a pricing strategy that will assist us in offering a low-cost solution that’s attainable for almost any commercial or public pool,” adds Bar-Ilan. “We’re aware that our analytics offering around various pool metrics and swimmer behaviour is a unique selling point, and we’ll continue to work on this development alongside the product’s fundamental drowning prevention capability.” “By providing an alert within the first critical 20 seconds of a drowning event, LYNXIGHT supports lifeguards in making better decisions and increasing their attention spans when early water distress signals are apparent,” concludes Bar-Ilan. “Sadly, drowning cases usually involve children, but the majority can be completely avoided by vigilant monitoring, and accurate and timely alerts.”
Keywords
LYNXIGHT, drowning, pool, lifeguard, swimmer, safety
